Cytochromes c555 from the Hyperthermophilic ... - ACS Publications

Biochemistry 2001, 40, 13681-13689

13681

Cytochromes c555 from the Hyperthermophilic Bacterium Aquifex aeolicus (VF5). 1. Characterization of Two Highly Homologous, Soluble and Membranous, Cytochromes c555† Frauke Baymann,‡ Pascale Tron,‡ Barbara Schoepp-Cothenet,*,‡ Corinne Aubert,‡ Pierre Bianco,‡ Karl-Otto Stetter,§ Wolfgang Nitschke,‡ and Michael Schu¨tz‡,| Laboratoire de Bioe´ nerge´ tique et Inge´ nierie des Prote´ ines (UPR9036), CNRS, 31 chemin Joseph Aiguier, F-13402 Marseille Cedex 20, France, and Lehrstuhl fu¨ r Mikrobiologie und Archaeenzentrum, UniVersita¨ t Regensburg, UniVersita¨ tsstrasse 31, D-93040 Regensburg, Germany ReceiVed June 11, 2001; ReVised Manuscript ReceiVed August 13, 2001

ABSTRACT: Two distinct class I (monoheme) c-type cytochromes from the hyperthermophilic bacterium Aquifex aeolicus were studied by biochemical and biophysical methods (i.e., optical and EPR spectroscopy, electrochemistry). The sequences of these two heme proteins (encoded by the cycB1 and cycB2 genes) are close to identical (85% identity in the common part of the protein) apart from the presence of an N-terminal stretch of 62 amino acid residues present only in the cycB1 gene. A soluble cytochrome was purified and identified by N-terminal sequencing as the cycB2 gene product. It showed an R-peak at 555 nm, an Em value of +220 mV, and electron paramagnetic resonance parameters of gz ) 2.89, gy ) 2.287, and gx ) 1.52. A firmly membrane-bound cytochrome characterized by nearly identical properties was detected and attributed to the cycB1 gene product. The very high degree of homology of its N-terminal part to cytochrome c553 from Heliobacterium gestii strongly suggests it to be anchored to the membrane via N-terminally attached lipid molecules. The two heme proteins were named cytochrome c555s (soluble) and cytochrome c555m (membranous). Electron paramagnetic resonance on partially ordered membrane multilayers suggests that the solvent-exposed heme domain of cytochrome c555m is flexible with respect to the membrane plane. Possible functional roles for both cytochromes are discussed.

The hyperthermophilic chemotroph Aquifex aeolicus (1) is a member of the Aquificales, considered to represent the earliest branching order of the phylogenetic tree of the Bacteria (2). A. aeolicus has been the target of a genome sequencing project achieved in 1998 (3) and has then entered the early proteomics era marked by an abundance of genomic information confronted by a scarcity of data with respect to metabolic mechanisms and functional/structural properties of enzymes involved. Recently, several components of the electron transfer chain in A. aeolicus have been studied. The structure of a soluble ferredoxin has been solved (4), the sulfide:quinone oxidoreductase (SQR) has been characterized in membranes (5), the [Ni/Fe] hydrogenases have been purified and studied with respect to functional properties and phylogenetic positioning (Guiral et al., unpublished data), and analysis of the cytochrome bc complex in membranes and partially purified isolates is underway (Schu¨tz et al., unpublished data). A detailed phylogenetic analysis of cytochrome bc-type † The work of K.-O.S. was financially supported by the Fonds der Chemischen Industrie, and the project on Aquifex at the BIP in Marseille benefited from financial support by the PCV program of the CNRS. * To whom correspondence should be addressed. Tel: +33 4 91164672. Fax: +33 4 91164578. E-mail: [email protected]. ‡ Laboratoire de Bioe ´nerge´tique et Inge´nierie des Prote´ines (UPR9036), CNRS. § Lehrstuhl fu ¨ r Mikrobiologie und Archaeenzentrum, Universita¨t Regensburg. | Present address: PROFOS AG, Regensburg, Germany.

enzymes (6) has recently shown that the enzyme from Aquifex is phylogenetically closely related to its homologue in the -proteobacteria Helicobacter pylori and Campylobacter jejuni, reminiscent of reports on a number of other proteins from this organism which were also found to be close to their proteobacterial counterparts with respect to amino acid sequence (7). On the basis of an inspection of the genomic environment of the genes coding for the bc complex in Aquifex, it was concluded that this species most probably has experienced extensive donation of genes from an -proteobacterium (6). A more detailed analysis of the open reading frames in the Aquifex genome indicated that about 20% of identifiable genes are close to -proteobacterial homologues, whereas the remaining 80% are strongly related to proteins from Thermotoga, Deinococcus, Thermus, and Archaea (Brugna and Nitschke, unpublished data), i.e., in line with the positioning of the species by 16S rRNA (8) and whole genome analysis (9). In this work, we describe the biochemical and physicochemical properties of a soluble and a membrane-attached cytochrome, highly homologous to each other. We compare the amino acid sequences of both cytochromes to those of homologous proteins from other species and discuss their possible functional roles. MATERIALS AND METHODS Bacterial Culture Conditions, Membrane Preparation, and Cytochrome Purification. A. aeolicus (VF5) was grown, and

10.1021/bi011201y CCC: $20.00 © 2001 American Chemical Society Published on Web 10/13/2001

13682 Biochemistry, Vol. 40, No. 45, 2001 cells were harvested as described previously (1). Membranes were prepared according to Nu¨bel et al. (5). When indicated, the membrane fraction was ultrasonicated in the presence of 2 M sodium bromide, stirred on ice for 1 h, and centrifuged for 1 h at 200000g in order to eliminate weakly membrane-attached proteins. For the purification of soluble cytochrome c555, cells were resuspended in 50 mM Hepes/ NaOH, pH 7, 4 mM MgCl2, 0.5 mM PMSF,1 and 10 µg/mL DNase. The soluble fraction was loaded on a DEAE-cellulose column equilibrated with a buffer containing 50 mM Hepes/ NaOH, pH 7, 5 mM EDTA, and 0.5 mM PMSF. The unretained fraction was subsequently applied onto a hydroxylapatite Bio-Gel column, and cytochrome c555s was eluted at 50 mM sodium phosphate buffer, pH 7.0. All purification steps were performed at room temperature. Optical and Electron Paramagnetic Resonance Spectroscopy. Optical spectra were recorded on an Aminco DW-2 spectrophotometer (SLM Instruments Inc., Urbana, IL); electron paramagnetic resonance (EPR) spectra were obtained using a Bruker ESP300e X-band spectrometer fitted with an Oxford Instrument He-cryostat and temperature control system. EPR redox titrations were performed as described by Dutton (10) in a buffer containing 50 mM MOPS, pH 7.0, 2 mM EDTA, and 40 mM NaCl in the presence of the following mediators at 100 µM: N,N,N′,N′-tetramethylp-phenylenediamine (TMPD), dichlorophenolindophenol (DCPIP), 2,5-dimethyl-p-benzoquinone, 2-hydroxy-1,2-naphthoquinone, cresyl blue, methylene blue, 2,5-dihydroxy-pbenzoquinone, indigo carmine, anthraquinone-2,4-disulfonate, anthraquinone-2-sulfonate, safranine T, and neutral red. Ferricyanide was present at 10 µM. Reductive titrations were carried out using sodium dithionite, and oxidative titrations were carried out using ferricyanide and potassium hexachloroiridate(IV). Optical redox titrations of purified soluble cytochrome c555s at 20 and 60 °C were performed electrochemically without addition of mediators as described previously (11). Oriented membrane multilayers were produced by partial dehydration in a humidity-controlled atmosphere (12, 13). Electrochemical Technique. Cyclic (CV) and square-wave (SWV) voltammetries (14) with a membrane-working electrode (15) were carried out using an EG&G 273A potentiostat controlled by EG&G PAR M 270/250 software. The CV scan rate generally was 20 mV s-1. SWV curves were obtained using 5 Hz as the square-wave frequency, 2 mV as the scan increment, and 25 mV as the pulse height amplitude. A conventional three-electrode system was used in 20 mM deoxygenated phosphate buffer, pH 7. Measurements at room temperature (23 °C) were performed using a polished basal plane pyrolytic graphite electrode. Experiments at higher temperatures required the use of gold electrodes pretreated for 1 min with 1 mM bis(4-pyridyl) disulfide solution (as proposed in ref 16). The reference electrode was a Metrohm Ag/AgCl/saturated NaCl electrode. The auxiliary electrode was a gold wire. The pH dependence of the redox midpoint potential was studied in a mixture of three buffer solutions composed of 1 Abbreviations: DEAE, diethylaminoethyl; EDTA, ethylenediaminetetraacetic acid; Em, redox midpoint potential; Em,7, Em at pH 7.0; EPR, electron paramagnetic resonance; MOPS, 3′-(N-morpholino)propanesulfonic acid; PMSF, phenylmethanesulfonyl fluoride; SDS-PAGE, sodium dodecyl sulfate-polyacrylamide gel electrophoresis.

Baymann et al.

FIGURE 1: Optical absorption difference (ascorbate reduced minus oxidized) spectra of the two cytochromes c555 from A. aeolicus recorded on the soluble fraction (representing cytochrome c555s, a) and on membrane fragments (representing cytochrome c555m, b).

10 mM sodium acetate/10 mM Tris chloride/10 mM sodium borate adjusted to the required pH. The temperature dependence of the redox midpoint potential was studied in the nonisothermal configuration as proposed in ref 17. Electrophoretic Analysis. SDS-PAGE was carried out according to the method of Laemmli (18). The gels were stained either with Coomassie Blue G-250 or with 3,3′,5,5′tetramethylbenzidine (TMBZ) as described by Thomas et al. (19). N-Terminal protein sequencing was carried out on the apoprotein with an Applied Biosystems gas-phase sequenator (models 470A and 473A). Sequence Analysis and Structural Models. Database searches for amino acid sequences were performed by BLASTP (20) with the amino acid sequence database at the National Center for Biotechnology Information, Washington, DC, or the TIGR microbiology database. The amino acid sequences were aligned with the help of the program CLUSTAL (21). Structural models of CycB2 from A. aeolicus, using cytochrome c552 from Thermus thermophilus as template, were obtained using Swiss-Model (22). All chemicals were of reagent grade and were purchased from commercial sources. RESULTS Identification of the Cytochromes c555 CycB2 was purified to homogeneity from the soluble fraction of disrupted A. aeolicus cells. Absorption difference spectra of the ascorbate-reduced cytochrome are shown in Figure 1, curve a. At 20 °C, reduced cytochrome CycB2’s R-band is split, with maxima at 555 and 550 nm. At variance with these spectral data, CycB2 is annotated as cytochrome c552 in the Aquifex genome. We therefore propose to rename CycB2 as cytochrome c555s (“s” standing for “soluble”). SDS-PAGE of the purified cytochrome, after staining with Coomassie Blue as well as with TMBZ (data not shown), shows a single band at approximately 9 kDa revealing a cytochrome with covalently attached heme as is typical for c-type cytochromes. The N-terminal peptide sequence of the heme-staining band was determined (ADGKAIFQQK). A BLAST search of this sequence in the A. aeolicus genome (3) (using the NCBI-genome server)

Characterization of Cyts c555s and c555m of A. aeolicus

Biochemistry, Vol. 40, No. 45, 2001 13683

FIGURE 2: Sequence alignment of cytochrome c555s (Accession Number: F70434; GI7430463) and c555m (B70369; GI7430462) from A. aeolicus, cytochrome c553 (AAC16774) from H. gestii, cytochrome c552 (S32485; GI479146) from H. thermophilus, cytochrome c552 (CCPS5S; GI65547) from Pseudomonas stutzeri, and cytochrome c554 (AAB88580; GI155079) from T. thermophilus. The heme binding site, the (putative) methionine ligand, and the histidine ligand are shown in bold. The vertical line indicates the start of the mature proteins. Residues conserved between the heliobacterial cytochrome c553 and c555m from Aquifex (Mb) and between all of the soluble proteins (Sol) are shown. Italic letters in cytochrome c555m emphasize its high degree of identical residues with respect to cytochrome c555s. Shading indicates R-helical structures (dark gray) and β-sheets (light gray) in the proteins either taken from the structure [TheTh (PDB entry: 1C52), PseSt (PDB entry: 1CCH), HydTh (PDB entry: 1AYG)] or predicted in the modeled structure of cytochrome c555s (AquAe).

showed that the N-terminal fragment corresponds to positions 18-27 of the cycB2 gene (Figure 2). CycB2 consists of 104 amino acid residues with a theoretical molecular weight of 11 600. The heme attachment site CXXCH, typical for c-type cytochromes, is located between positions 29 and 32. Two methionine residues, Met-78 and Met-84, are present in the sequence stretch that commonly contains the sixth Met ligand for the heme iron in class I cytochromes (see below). The N-terminal stretch of 17 amino acid residues of the cycB2 gene product, absent in the mature protein, displays the typical features of a signal peptide for translocation across the cytoplasmic membrane into the periplasmic space, i.e., a positively charged N-terminus followed by a hydrophobic region and a signal peptidase cleavage site (23). Cleavage of the N-terminal signal peptide in the mature protein argues for a periplasmic localization of this cytochrome. A molecular weight of 9901 is expected for the mature protein in agreement with the apparent molecular weight of CycB2 on SDS-PAGE. Sequence comparisons to a variety of small periplasmic soluble cytochromes revealed significant homology (identity >30%, homology >50%) to cytochromes from other Aquificales (cytochrome c552 from Hydrogenobacter thermophilus), from the Deinococcus/Thermus group (cytochrome c554 from T. thermophilus), and from Firmicutes (cytochrome c553 from Heliobacillus gestii) as well as from proteobacteria (cytochrome c552 from Nitrosomonas europaea and cytochrome c551 of various Pseudomonas strains) (Figure 2). The three-dimensional structures of these cytochromes (for those known) were found to be rather similar to each other (24-27). The structure of cytochrome c555 from Aquifex can thus be modeled using cytochrome c554 from Thermus

as template, yielding a similar overall structure for the Aquifex cytochrome. Interestingly, no significant conservation is observed in the region around the two methionine residues. The closest sequence neighbor to CycB2 from Aquifex is none of the above-mentioned soluble cytochromes but rather a second cytochrome from the same species, i.e., the cycB1 gene product. CycB2 is in fact 85% identical and 89% homologous to the last two-thirds of the carboxy-terminal end of CycB1 (Figure 2). The cycB1 gene predicts a protein of 166 amino acid residues with a theoretical molecular weight of 19 135. Properties of Cytochrome c555s pH Dependence of the Midpoint Redox Potential. A redox midpoint potential of +220 mV was determined by electrochemical redox titration at pH 7 and 20 °C (Figure 3). The pH dependence of the redox midpoint potential of cytochrome c555s, measured from CV and SWV data, is depicted in Figure 4. The amplitudes of the CV and SWV signals decreased when the pH was increased from 8 to 11 or was decreased from 6 to 3. Signal sizes were partially restored when the pH was raised back or lowered back from 3 or 11 to neutrality, respectively (data not shown). Redox midpoint potential values remained roughly constant in the pH range of 6-8. Above pH 8 and below pH 6, the redox midpoint potential was pH dependent. The alkaline pH dependence was characterized by a pK value on the oxidized form of the heme (pKox2) of 8.4 whereas the acidic region of the Em vs pH curve could be fitted assuming pK values of 4 (pKox1) and 5.2 (pKred).

13684 Biochemistry, Vol. 40, No. 45, 2001

FIGURE 3: Potentiometric titration of the soluble cytochrome c555s from A. aeolicus. Electrochemical titration was performed as described in Moss et al. (11) without addition of mediators in a buffer containing 50 mM MOPS, pH 7.0, and 100 mM KCl. Spectral amplitude variation was followed in the range of 400 to +600 nm. The data were fitted by an n ) 1 Nernst curve with a midpoint potential of +220 mV. Reductive and oxidative titrations were superimposable.

FIGURE 4: pH dependence of the redox midpoint potential of cytochrome c555s from A. aeolicus. A buffer mixture of 10 mM sodium acetate/10 mM Tris chloride/10 mM sodium borate was used, and the pH was adjusted to the values indicated. Redox midpoint potentials were determined by both cyclic and squarewave voltammetries as described under Materials and Methods. The obtained results were fitted by the equation (57) Em ) Eˆ - 0.06 log[([H+]2 + Kox1[H+] + Kox1Kox2)/([H+]2 +Kred[H+])], where [H+] is the proton concentration in solution, Eˆ is the midpoint potential of the fully protonated form, and pKox1 (4 ( 0.4), pKred (5.2 ( 0.2), and pKox2 (8.4 ( 0.1) are three pK values of oxidized and reduced forms. The inset represents optical spectra recorded on cytochrome c555s at pH 7 (straight line) and at pH 10 (dashed line).

pKox values above 8 in class I cytochromes are frequently associated to loss of methionine coordination of the heme iron (28). Optical spectra recorded on oxidized cytochrome c555s at pH 7 and 10 indeed demonstrated a loss of the 695 nm absorption band, commonly attributed to a sulfur iron charge transfer band, i.e., indicative of methionine ligation (inset in Figure 4). This suggests that, like many other class I cytochromes, cytochrome c555s undergoes ligand changes at high pH values. The pKox1 and pKred values (4 and 5.2, respectively), characterizing the pH dependence beyond neutrality, are separated by approximately one pH unit. A similar pH

Baymann et al.

FIGURE 5: Temperature dependence of the redox midpoint potential. Redox midpoint potentials were determined, as described in Materials and Methods, by cyclic and square-wave voltammetries. A straight line characterized by a temperature coefficient dE°′/dT of -0.2 mV‚K-1 corresponding to a standard entropy change S°′ of -19 J‚mol-1‚K-1 was fitted to the data points. Spectra recorded on cytochrome c555s at 20 °C (broad line) and 60 °C (dashed line) as well as the difference between these two spectra (dotted line) are shown in the inset.

dependence of the redox midpoint potential was determined for the Rhodomicrobium Vannielii and Rhodospirillum rubrum cytochromes c2 (28). The pKox1 and pKred values are in these cases associated to the ionization of the rear heme propionate. The precise values of pKox and pKred are proposed to depend on the nature of the residues interacting with the propionate oxygens (28). Temperature Dependence of the Redox Midpoint Potential. Two series of measurements were performed using CV and SWV techniques, either with or without renewing the sample at each investigated temperature. Very similar results were gained from both series of experiments. The temperature dependence of Em,7 is shown in Figure 5. The obtained curve yields a temperature coefficient dEm,7/dT of -0.2 mV‚K-1. The split R-band observed at 20 °C changed its form when temperature was increased (inset in Figure 5), both absorption bands shifting and overlapping. EPR Spectroscopy. The EPR spectrum of the oxidized form of cytochrome c555s was characteristic of a low-spin heme with signals at gz ) 2.89, gy ) 2.287, and gx ) 1.52 (Figure 6, panel a). Properties of Membrane-Bound Cytochrome c555 Optical and EPR spectra recorded on membranes from A. aeolicus demonstrated the presence of a cytochrome with optical (spectrum b in Figure 1) and EPR (panel b in Figure 6) parameters similar to those of cytochrome c555s. Both the optical and EPR spectra of this cytochrome persisted even in membranes treated with chaotropic agent. This suggested that the respective cytochrome was indeed tightly attached to the membrane and was distinct from cytochrome c555s. We therefore refer to this heme protein in the following as cytochrome c555m. Redox titration of the gz ) 2.88 EPR signal of cytochrome c555m in membrane fragments yielded a midpoint redox potential of +210 ( 20 mV (n ) 1) (Figure 7).

Characterization of Cyts c555s and c555m of A. aeolicus

FIGURE 6: EPR spectra of the soluble (panel a) and the membraneattached (panel b) c555 cytochromes. Spectra were recorded on samples without addition (straight line) or treated with 5 mM ascorbate (dashed line). gx, gy, and gz values for cytochrome c555s were determined at 1.52, 2.287, and 2.89, respectively, whereas they were found at 1.53, 2.296, and 2.88, respectively, for the membrane-attached cytochrome c555m.

FIGURE 7: EPR titration of the membrane-attached cytochrome c555m. Titrations were performed in a buffer containing 50 mM MOPS, pH 7.0, 2 mM EDTA, and 40 mM NaCl in the presence of redox mediators as indicated in Materials and Methods. The titration curve represents the dependence of the gz ) 2.88 signal size on ambient redox potential. The data points were fitted by an n ) 1 Nernst curve assuming a redox midpoint potential of +210 mV.

EPR Experiments on Partially Ordered Membrane Fragments. EPR spectra were recorded on oriented, untreated (i.e., largely oxidized) membranes of A. aeolicus in a range of angles between -80° to +90° (magnetic field with respect to the plane of the membrane) (Figure 8a). The sample was well ordered as judged from the strong anisotropy of the gz peak at g ) 3.03, attributed to heme a of the cytochrome aa3 oxidase (to be published) (Figure 8b, open circles). In

Biochemistry, Vol. 40, No. 45, 2001 13685

FIGURE 8: EPR study of two-dimensionally ordered membrane fragments. Panel a: EPR spectra recorded on oriented membrane multilayers prepared according to Rutherford and Se´tif (13) without adding exogenous oxidant or reductant and exposed to air. Under these conditions, redox centers with Em values below +300 mV were found to be largely oxidized. The magnetic field was oriented 90° (straight line) or 0° (dashed line) with respect to the plane of the membrane. The peak at g ) 2.88 corresponds to the gz line of cytochrome c555m and that at g ) 3.03 is attributed to cytochrome oxidase. Panel b: Polar plot evaluation of the signal amplitudes of the gz peaks arising from cytochrome oxidase (open circles) and from cytochrome c555m (closed circles) as well as the gx and gy lines of cytochrome c555m.

contrast to this signal (and several others that are not shown), the orientation dependence of the gz ) 2.88 line intensity (attributed to cytochrome c555m) showed multiple maxima. A polar plot evaluation of the gz ) 2.88 signal amplitude (closed circles in Figure 8b) indicates three weak maxima at 30°, 50°, and 80° with respect to the membrane. The polar plot evaluation of the intensity of the gx ) 1.53 trough, although less accurate due to lower signal-to-noise ratio, also indicated several maxima. DISCUSSION Phylogenetic Analysis. BLAST searches in sequence databases for homologous proteins yielded predominantly class I c-type cytochromes, such as the small soluble cytochromes from several proteobacteria, from the Aquificalis H. thermophilus, or from T. thermophilus as well as the membrane-attached cytochrome c553 from H. gestii. A multiple alignment of the retrieved sequences was performed with the help of CLUSTAL X, and phylogenetic trees were calculated using the Neighbor-Joining method. It is noteworthy that the N-terminal extension of the membraneattached heme proteins cytochrome c555m from A. aeolicus and cytochrome c553 from H. gestii was not taken into account

13686 Biochemistry, Vol. 40, No. 45, 2001 in this analysis. The obtained trees varied slightly in topological details, presumably due to the restricted number of sites in these very small proteins. A few aspects, however, were common to all calculated trees and are therefore worth mentioning. (i) Both Aquifex cytochromes form a clade (bootstrap of 100) showing that the presence of two cytochromes c555 is due to gene duplication in the Aquificales. (ii) The A. aeolicus cytochromes c555 cluster with the heliobacterial cytochrome c553 (it is noteworthy that this does not result from a bias conferred by the N-terminal extension which was omitted from the analysis). The substantial sequence homology of the N-terminal extension (Figure 2) together with this phylogenetic proximity of the C-terminal heme domains indicates that the parent gene in A. aeolicus was probably that of the membrane-attached cytochrome c555m. Gene duplication and N-terminal truncation may then have resulted in the soluble cytochrome c555s. (iii) Cytochrome c555 from A. aeolicus, cytochrome c553 from H. gestii, and cytochrome c554 from T. thermophilus form a clade which is well separated from that containing the proteobacterial cytochromes and cytochrome c552 from H. thermophilus. There are several reports in the literature showing that A. aeolicus contains a substantial number of genes related to proteobacterial ones (7), and it was proposed that Aquifex has experienced extensive lateral gene transfer from a proteobacterial donor (6). A recent analysis suggests that about 20% of the A. aeolicus genome represents genes related to proteobacteria (Brugna and Nitschke, unpublished). Since the Aquifex cytochromes do not cluster together with the proteobacterial counterparts, these two heme proteins may in fact be part of the genuine heritage of the Aquificales lineage. (iv) The Hydrogenobacter cytochrome, by contrast, was found to cluster with the proteobacterial proteins in all trees obtained. This would indicate that the Aquificalis H. thermophilus uses a proteobacterial gene to code for its soluble cytochrome whereas A. aeolicus employs a lineagerelated protein. In the model of lateral gene transfer into the Aquificales mentioned above, this would mean that either H. thermophilus has acquired its cytochrome c552 gene after the branching off of A. aeolicus or that the common ancestor of H. thermophilus and A. aeolicus contained both lines of cytochromes, one or the other of which was lost in the descendent species. Adaptation to Hyperthermophilic Conditions. The determination of the factors leading to thermostability represents a major goal in the study of thermo- and hyperthermophilic organisms. For the particular case of a mesophilic representative of the small soluble class I c-type cytochromes, sitedirected mutagenesis guided by sequence and structure comparisons with its thermophilic homologue in the Aquificalis H. thermophilus has recently succeeded in confering thermostability to this cytochrome, thereby showing that the modification of only a few selected residues can result in thermostability (29). As mentioned above, the Hydrogenobacter cytochrome appears to be phylogenetically closely related to its proteobacterial counterparts. A straightforward comparison between these proteins was therefore possible. The equivalent approach is not possible for the case of both Aquifex cytochromes c555. The determination of the 3D structure of at least one of these two cytochromes and a subsequent comparison of sequence and structural characteristics between the proteobacterial and the Hydrogenobacter

Baymann et al. cytochromes on one side and the Thermus and Aquifex cytochromes on the other side may therefore provide more profound insight into the parameters distinguishing between meso- (80 °C) proteins. Overproduction of the A. aeolicus cytochromes in E. coli has therefore been initiated [see accompanying article (58)] as a crucial prerequisite to ultimately determine the structure. In the native soluble cytochrome c555s from A. aeolicus, the dependence of the redox midpoint potential on temperature was examined. The Em was found to be only weakly dependent on T, amounting to a total change of 12 mV in the range from 20 to 80 °C. The deduced entropy contribution turned out to be relatively low as compared to that of mesoand thermophilic proteins (17, 30, 31). The constant slope of the Em vs T curve in the region explored furthermore showed that no major conformational changes occur in this temperature range. Thermostability parameters of the heterologously expressed protein and mutants thereof are described in the accompanying article (58). Two Possible Candidates for the Sixth Ligand Methionine. The positioning of the c-heme attachment motif (CXXCH) in the N-terminal region of the sequence and of the methionine residues putatively serving as the sixth heme ligand approximately 40 residues further toward the Cterminal end suggested that both cytochromes c555 belong to the class I c-type cytochromes. Following Ambler’s classification scheme (32), the presence of a M...LS...I motif downstream of the methionine ligand as well as of a split R-band defines the cytochromes c555 as representatives of the subclass Ic. Unlike most class I cytochromes, two different methionine residues (five and six residues apart in the cytochrome c555m and cytochrome c555s sequences, respectively) are present in the sequence stretch typically harboring the sixth ligand methionine. In proteobacterial class I cytochromes, the second residue after the ligating methionine frequently is a proline. On the basis of the presence of such a proline following the first of the two methionines in both Aquifex cytochromes, this first methionine was aligned with the sixth ligand of the proteobacterial proteins. Interestingly, however, two methionines, six residues apart, are also present in cytochrome c554 from T. thermophilus. The crystal structure of the Thermus cytochrome has been solved (27), showing that the second of the two Met residues is the axial ligand. This second methionine was consequently aligned with the ligating Met of the class I proteobacterial cytochromes in Figure 2. Since homology in this sequence region between the proteins from Aquifex, Thermus, and the proteobacteria is low, an alignment of the two respective Met residues of Aquifex and Thermus is equally possible. Such an alignment would suggest the second Met residue in the Aquifex proteins as the heme ligand. Sequence comparisons alone therefore did not allow us to unequivocally identify the methionine serving as the sixth ligand in the A. aeolicus cytochromes. This question was therefore addressed by site-directed mutagenesis, and the respective results are presented in the accompanying article (58). Cellular Localization of Both Cytochromes and Mode of Membrane Attachment of Cytochrome c555m. The N-terminal sequence of the mature form of cytochrome c555s was determined, demonstrating that the first 17 amino acid

Characterization of Cyts c555s and c555m of A. aeolicus residues of the gene product represent a cleaved-off leader peptide. The characteristics of this leader peptide suggested that it serves in translocation of cytochrome c555s into the periplasmic space. The sequence of the 17-residue leader peptide of cytochrome c555s is strongly conserved in the gene sequence of cytochrome c555m, suggesting that also this latter heme protein is addressed to the periplasmic space. In contrast to its soluble counterpart, cytochrome c555m fractionated with the membrane and even treatment of membranes with chaotropic salts did not liberate any soluble form of cytochrome c555m, indicating a strong association of this heme protein to the lipid bilayer. This raised the question of the mode of anchoring of cytochrome c555m to the membrane. The astonishingly high homology of the C-terminal part of cytochrome c555m’s amino acid sequence to that of its soluble counterpart suggests that membrane attachment is mediated by the N-terminal extension of cytochrome c555m. Apart from the supposed leader peptide, no hydrophobic stretch could be identified in this N-terminal extension. A clue to the probable mode of membrane attachment came from the significant sequence homology of cytochrome c555m to cytochrome c553 from H. gestii. Heliobacterial cytochrome c553 has been shown to be a lipoprotein with the lipid residues presumably attached to an N-terminal Cys residue in the mature protein (33). Intriguingly, a Cys residue is present in cytochrome c555m from A. aeolicus immediately following the first 17 amino acid residues supposed to be a leader peptide. These residues including the Cys aligned very well with the residues in H. gestii cytochrome c553. We therefore propose that cytochrome c555m from Aquifex also is a lipoprotein. Confirmation of this hypothesis and a more detailed characterization of the lipid molecules require purification of sufficient quantities of cytochrome c555m from the membrane fraction, which has not been achieved so far. If this model is correct, A. aeolicus cytochrome c555m represents one more exception (see also refs 33 and 34) to the previously stipulated rule of lipoproteins being characterized by a LAGC/LAAC signal peptide sequence necessary for binding of the signal peptidase (23). Cytochrome c555m Belongs to the Family of PiVoting Cytochromes. The described EPR results obtained on partially ordered membrane multilayers showed that the cytochrome c555m heme can adopt several (or possibly even a continuum of) orientations with respect to the membrane. This indicated that the membrane extrinsic domain of cytochrome c555m performs pivoting movements on the membrane surface. The obtained results are reminiscent of those obtained by EPR on the Rieske protein in cytochrome bc complexes (35-38) and in particular of those described for another membrane-attached cytochrome, i.e., cytochrome b558/566 from the Archaeon Sulfolobus acidocaldarius (39). Whereas for the case of the Sulfolobus cytochrome b558/566 only the gz value was detected, the orientations of all three principal g-tensor directions could be observed for cytochrome c555m. Multiple maxima were seen for gz whereas the gy direction displayed a single and well-defined maximum parallel to the membrane plane. In the limit of the poor signal-to-noise ratio, the gx signal also appeared to show more than one maximum. A straightforward rationalization of these observations consists of assuming that the gy axis is close to the rotation axis of the protein and that the conformational motion would therefore be restricted and guided rather than

Biochemistry, Vol. 40, No. 45, 2001 13687 represent a mere “dangling” of the extrinsic domain from the membrane surface. At the present time, however, this interpretation should be taken with caution. Whereas for paramagnetic centers with a fixed conformation the observed g directions always form an orthonormal set (40), this rule does not seem to hold for the case of “mobile-domain” centers (36, 38). A possible explanation for this phenomenon consists of assuming variable population densities along the different g axes. For instance, a given conformational substate may be well-ordered along one g direction, correspondingly leading to an intense EPR signal, but relatively disordered along one or both of the remaining directions, resulting in weak signals possibly difficult to distinguish in a polar plot. A quantitative evaluation of the polar plot dependences in these wobbling centers is therefore not possible at this time. Computational simulations of the experimental polar plots might help to remediate this problem, and we are therefore presently initiating efforts in this direction. The possibility of domain movement in membraneattached cytochromes was evoked as early as 1979 for the case of liver microsomal cytochrome b5 (41). More recently, sequence and kinetic data on cytochrome cy from Rhodobacter capsulatus indicated a respective flexibility for this heme protein (42-44). Close relatives of cytochrome cy have subsequently been identified in several proteobacterial species (45-47). Kinetic arguments in favor of the possibility of domain movements have also been reported for cytochrome cz in green sulfur bacteria (48). A comparable domain flexibility was proposed for membrane-attached cytochromes in Bradyrizobia (45) and Gram-positive bacteria (49). In the case of the cy cytochromes, the extrinsic heme binding domain is linked to the membrane anchor via a stretch of 43-61 amino acid residues (45-47, 50). In A. aeolicus cytochrome c555m, 62 residues connect the domain homologous to cytochrome c555s to the presumable Nterminal lipid anchor. Current prediction algorithms did not yield an evident secondary structure for this sequence stretch, and it can therefore not be decided at present whether this stretch merely acts as a leash or whether it is folded into a specific structure. Several of the mentioned cytochromes are parts of larger enzyme supercomplexes. The cytochromes c552 from Paracoccus denitrificans and c551 from Bacillus PS3, for example, are integrated into a supercomplex additionally containing the cytochrome bc complex and cytochrome oxidase (49, 51). At present, we do not know whether cytochrome c555m represents an isolated membrane-attached heme protein or whether it is part of a larger entity. Purification attempts are in progress to address this question. Functional Role of the Two Cytochromes c555. No direct evidence concerning the functional roles of cytochrome c555s and cytochrome c555m in A. aeolicus has been obtained so far. Their redox midpoint potentials suggest that one or both might be involved in electron transfer from the cytochrome bc complex toward terminal oxidases similar to the task fulfilled by cytochrome cy in proteobacteria. In such a scenario, the two different cytochromes may either be specific to different electron transfer pathways or act as competing or successive carriers in the same chain. In the case of R. capsulatus for example, the soluble and membranebound counterparts have both competing functions in the cytochrome bc1/aa3 oxidase chain (44, 52) but different

13688 Biochemistry, Vol. 40, No. 45, 2001 specific functions in the nitrous and nitric oxide reductase activity, respectively (53, 54). In this context, it seems noteworthy to us that none of the other species studied so far contains two cytochromes with so similar properties (with respect to sequence, redox midpoint potential, and optical and EPR spectra). Whereas this may admittedly represent a rather recent gene duplication, it could also reflect functional constraints requiring conservation of the extrinsic domain. Cytochrome c555s has a tendency to associate into multimeric forms influencing both redox and optical properties (data not shown). The similarity between the heme binding domains of cytochrome c555m and cytochrome c555s may suggest interaction between the two proteins. Such a heterologous association could create an electron transfer couple as reported for the diheme cytochrome c4 (55) or for the dimeric cytochrome c552 from Pseudomonas nautica (56). A dynamic analysis of the Aquifex electron transfer chain and the identification of possible supercomplex formation between the enzymes involved may help to further elucidate the detailed functional role of the two c555 cytochromes. ACKNOWLEDGMENT We thank R. Huber (Regensburg, FRG) for providing cell material of Aquifex aeolicus. We also thank the group of P. Bertrand (Marseille, France) for extensive access to their EPR equipment. REFERENCES 1. Huber, R., Wilham, T., Huber, D., Trincone, A., Burggraf, S., Ko¨nig, H., Rachel, R., Rockinger, I., Fricke, H., and Stetter, K. O. (1992) Syst. Appl. Microbiol. 15, 340-351. 2. Olsen, G. J., Woese, C. R., and Overbeek, R. (1994) J. Bacteriol. 176, 1-6. 3. Deckert, G., Warren, P. V., Gaasterland, T., Young, W. G., Lenox, A. L., Graham, D. E., Overbeek, R., Snead, M. A., Keller, M., Aujay, M., Huber, R., Feldman, R. A, Short, J. M., Olsen, G. J., and Swanson, R. V. (1998) Nature 392, 353358. 4. Yeh, A. P., Chatelet, C., Soltis, S. M., Kuhn, P., Meyer, J. and Rees, D. C. (2000) J. Mol. Biol. 14, 587-595. 5. Nu¨bel, T., Klughammer, C., Huber, R., Hauska, G., and Schu¨tz, M. (2000) Arch. Microbiol. 173, 233-244. 6. Schu¨tz, M., Brugna, M., Lebrun, E., Baymann, F., Huber, R., Stetter, K.-O., Hauska, G., Toci, R., Lemesle-Meunier, D., Tron, P., Schmidt, C., and Nitschke, W. (2000) J. Mol. Biol. 300, 663-675. 7. Klenk, H. P., Meier, T. D., Durovic, P., Schwass, V., Lottspeich, F., Dennis, P. P., and Zillig, W. (1999) J. Mol. EVol. 48, 528-541. 8. Nelson, K. E., Clayton, R. A., Gill, S. R., Gwinn, M. L, Dodson, R. J., Haft, D. H., Hickey, E. K., Peterson, J. D., Nelson, W. C., Ketchum, K. A., McDonald, L., Utterback, T. R., Malek, J. A., Linher, K. D., Garrett, M. M., et al. (1999) Nature 399, 323-329. 9. Snel, B., Bork, P., and Huyen, M. A. (1999) Nat. Genet. 21, 108-110. 10. Dutton, P. L. (1971) Biochim. Biophys. Acta 226, 63-80. 11. Moss, D. A., Nabedryk, E., Breton, J., and Ma¨ntele, W. (1990) Eur. J. Biochem. 187, 565-572. 12. Blasie, J. K., Erecinska, M., Samuels, S., and Leigh, J. S. (1978) Biochim. Biophys. Acta 501, 33-52. 13. Rutherford, A. W., and Se´tif, P. (1990) Biochim. Biophys. Acta 1019, 128-132. 14. Osteryoung, J., and O’Dea, J. J. (1986) in Electroanalytical Chemistry: A Series of AdVances (Bard, A. J., Ed.) Vol. 14, pp 209-110, Marcel Dekker, New York.

Baymann et al. 15. Haladjian, J., Bianco, P., Nunzi, F., and Bruschi, M. (1994) Anal. Chim. Acta 289, 15-20. 16. Taniguchi, I., Toyosawa, K., Yamaguchi, H., and Yasukouchi, K. (1982) J. Chem. Soc., Chem. Commun., 1032. 17. Taniguchi, V. T., Sailasuta-Scott, N., Anson, F. C., and Gray, H. B. (1980) Pure Appl. Chem. 52, 2275. 18. Laemmli, U. K. (1970) Nature 227, 680-685. 19. Thomas, P. E., Ryan, D., and Levin, W. (1976) Anal. Biochem. 75, 168-176. 20. Altschul, S. F., Gish, W., Miller, W., Myers, E. W., and Lipman, D. J. (1990) J. Mol. Biol. 215, 403-410. 21. Thomson, J. D., Gilson, T. J., Plewniak, F., Jeanmougin, F., and Higgins, D. G. (1997) Nucleic Acids Res. 24, 4876-4882. 22. Guex, N., and Peitsch, M. C. (1997) Electrophoresis 18, 27142723. 23. von Heijne, G. (1988) Biochim. Biophys. Acta 947, 307-333. 24. Timkovich, R., Bergmann, D., Arciero, D. M., and Hooper, A. B. (1998) Biophys. J. 75, 1964-1972. 25. Cai, M., and Timkovich, R. (1994) Biophys. J. 67, 1207-1215. 26. Hasegawa, J., Yoshida, T., Yamazaki, T., Sambongi, Y., Yu, Y., Igarashi, Y., Kodama, T., Yamazaki, K.-I., Kyogoku, Y., and Kobayashi, Y. (1998) Biochemistry 37, 9641-9649. 27. Than, M. E., Hof, P., Huber, R., Bourenkov, G. P., Bartunik, H. D., Buse, G., and Soulimane, T. (1997) J. Mol. Biol. 271, 629-644. 28. Moore, G. R., and Pettigrew, G. W. (1990) in Cytochromes c. EVolutionary, structural and physicochemical aspects, SpringerVerlag, Berlin. 29. Hasegawa, J., Uchiyama, S., Tanimoto, Y., Mizutani, M., Kobayashi, Y., Sambongi, Y., and Igarashi, Y. (2000) J. Biol. Chem. 275, 37824-37828. 30. Verhagen, M. F. J. M., Wolbert, R. B. G., and Hagen, W. R. (1994) Eur. J. Biochem. 221, 821-829. 31. Florens, L., Bianco, P., Haladjian, J., Bruschi, M., Protasevich, I., and Makarov, A. (1995) FEBS Lett. 373, 280-284. 32. Ambler, R. P. (1991) Biochim. Biophys. Acta 1058, 42-47. 33. Albert, I., Rutherford, A. W., Grav, H., Kellermann, J., and Michel, H. (1998) Biochemistry 37, 9001-9008. 34. Weyer, K. A., Scha¨fer, M., Lottspeich, F., and Michel, H. (1987) Biochemistry 26, 2909-2914. 35. Brugna, M., Albouy, D., and Nitschke, W. (1998) J. Bacteriol. 180, 3719-3723. 36. Schoepp, B., Brugna, M., Riedel, A., Nitschke, W., and Kramer, D. (1999) FEBS Lett. 450, 245-251. 37. Brugna, M., Nitschke, W., Asso, M., Guigliarelli, B., LemesleMeunier, D., and Schmidt, C. (1999) J. Biol. Chem. 274, 16766-16772. 38. Brugna, M., Rodgers, S., Schriker, A., Montoya, G., Kazmeier, M., Nitschke, W., and Sinning, I. (2000) Proc. Natl. Acad. Sci. U.S.A. 97, 2069-2074. 39. Schoepp-Cothenet, B., Schu¨tz, M., Baymann, F., Brugna, M., Nitschke, W., Myllykallio, H., and Schmidt, C. (2001) FEBS Lett. 487, 372-376. 40. Hootkins, R., and Bearden, A. (1983) Biochim. Biophys. Acta 723, 16-29. 41. Rich, P. R., Tiede, D. M., and Bonner, W. D., Jr. (1979) Biochim. Biophys. Acta 546, 307-315. 42. Prince, R. C., Davidson, E., Haith, C. E., and Daldal, F. (1986) Biochemistry 25, 5208-5214. 43. Jenney, F. E., Prince, R. C., and Daldal, F. (1994) Biochemistry 33, 2496-2502. 44. Myllikallio, H., Drepper, F., Mathis, P., and Daldal, F. (1998) Biochemistry 37, 5501-5510. 45. Bott, M., Ritz, D., and Hennecke, H. (1991) J. Bacteriol. 173, 6766-6772. 46. Turba, A., Jetzek, M., and Ludwig, B. (1995) Eur. J. Biochem. 231, 259-265. 47. Myllykallio, H., Zannoni, D., and Daldal, F. (1999) Proc. Natl. Acad. Sci. U.S.A. 96, 4348-4353. 48. Oh-oka, H., Iwaki, M., and Itoh, S. (1997) Biochemistry 36, 9267-9272. 49. Sone, N., Sekimashi, M., and Kutoh, E. (1987) J. Biol. Chem. 262, 15386-15391.

Characterization of Cyts c555s and c555m of A. aeolicus 50. Myllykallio, H., Jenney, F. E., Moomaw, C. R., Slaughther, C. A., and Daldal, F. (1997) J. Bacteriol. 179, 2623-2631. 51. Berry, A. E, and Trumpower, B. L. (1985) J. Biol. Chem. 260, 2458-2467. 52. Hochkoeppler, A., Jenney, F. E., Lang, S. E., Zannoni, D., and Daldal, F. (1995) J. Bacteriol. 177, 608-613. 53. Richardson, D. J., Bell, L. C., MaEwan, A. G., Jackson, J. B., and Ferguson, S. J. (1991) Eur. J. Biochem. 199, 677-683. 54. Bell, L. C., Richardson, D. J., and Ferguson, S. J. (1992) J. Gen. Microbiol. 138, 437-443. 55. Kadziola, A., and Larsen, S. (1997) Structure 5, 203-216.

Biochemistry, Vol. 40, No. 45, 2001 13689 56. Brown, K., Nurizzo, D., Besson, S., Shepard, W., Moura, J., Moura, I., Tegoni, M., and Cambillau, C. (1999) J. Mol. Biol. 289, 1017-1028. 57. Clark, W. M. (1960) in Oxidation Reduction Potential of Organic Systems, Williams and Wilkins, Baltimore, MD. 58. Aubert, C., Guerlesquin, F., Bianco, P., Leroy, G., Tron, P., Stetter, K.-O., and Bruschi, M. (2001) Biochemistry 40, 13690-13698. BI011201Y